netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Gagan Arneja <gaagaan@gmail.com>
To: Sridhar Samudrala <sri@us.ibm.com>
Cc: Rick Jones <rick.jones2@hp.com>,
	Krishna Kumar2 <krkumar2@in.ibm.com>,
	Evgeniy Polyakov <johnpol@2ka.mipt.ru>,
	netdev@vger.kernel.org
Subject: Re: [RFC] New driver API to speed up small packets xmits
Date: Thu, 10 May 2007 12:43:53 -0700	[thread overview]
Message-ID: <464375F9.1050600@gmail.com> (raw)
In-Reply-To: <1178820423.17393.3.camel@w-sridhar2.beaverton.ibm.com>

> small packets belonging to the same connection could be coalesced by
> TCP, but this may help the case where multiple parallel connections are
> sending small packets.

It's not just small packets. The cost of calling hard_start_xmit/byte 
was rather high on your particular device. I've seen  PCI read 
transaction in hard_start_xmit taking ~10,000 cycles on one particular 
device. Count the cycles your brand of NIC is taking in it's 
xmit_routine. The worse it is, the stronger your case for cluster 
transmits.

Also, I think, you don't have to chain skbs, they're already chained in 
  Qdisc->q. All you have to do is take the whole q and try to shove it 
at the device hoping for better results. But then, if you have rather 
big backlog, you run the risk of reordering packets if you have to requeue.

> 
> Thanks
> Sridhar

--
Gagan

> 
> -
> To unsubscribe from this list: send the line "unsubscribe netdev" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

  reply	other threads:[~2007-05-10 19:43 UTC|newest]

Thread overview: 86+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-05-10 14:53 [RFC] New driver API to speed up small packets xmits Krishna Kumar
2007-05-10 15:08 ` Evgeniy Polyakov
2007-05-10 15:22   ` Krishna Kumar2
2007-05-10 15:48     ` Evgeniy Polyakov
2007-05-10 16:08       ` jamal
2007-05-10 17:19     ` Rick Jones
2007-05-10 18:07       ` Sridhar Samudrala
2007-05-10 19:43         ` Gagan Arneja [this message]
2007-05-10 20:11           ` jamal
2007-05-10 20:14             ` Rick Jones
2007-05-10 20:15               ` jamal
2007-05-10 20:15             ` Gagan Arneja
2007-05-10 20:21               ` jamal
2007-05-10 20:25                 ` Gagan Arneja
2007-05-11  5:22             ` Krishna Kumar2
2007-05-11 11:27               ` jamal
2007-05-10 20:37           ` David Miller
2007-05-10 20:40             ` Gagan Arneja
2007-05-10 20:57               ` David Miller
2007-05-11  6:07                 ` Krishna Kumar2
2007-05-11  5:21             ` Krishna Kumar2
2007-05-11  5:20           ` Krishna Kumar2
2007-05-11  5:35             ` Gagan Arneja
2007-05-11  5:43               ` Krishna Kumar2
2007-05-11  5:57                 ` Gagan Arneja
2007-05-11  6:06                   ` Krishna Kumar2
2007-05-11  6:29                     ` Gagan Arneja
2007-05-11  6:52                       ` Krishna Kumar2
2007-05-10 18:13       ` Vlad Yasevich
2007-05-10 18:20         ` Rick Jones
2007-05-10 18:32           ` Vlad Yasevich
2007-05-10 18:40             ` Rick Jones
2007-05-10 18:59             ` Ian McDonald
2007-05-10 19:21               ` Vlad Yasevich
2007-05-10 19:26                 ` Ian McDonald
2007-05-10 20:32                 ` David Miller
2007-05-10 20:49                   ` Rick Jones
2007-05-10 21:02                     ` David Miller
2007-05-10 21:14                       ` Gagan Arneja
2007-05-11  2:28                         ` Stephen Hemminger
2007-05-11  5:01                           ` Gagan Arneja
2007-05-11  5:04               ` Krishna Kumar2
2007-05-11  9:01                 ` Evgeniy Polyakov
2007-05-11  9:18                   ` Krishna Kumar2
2007-05-11  9:32                     ` Evgeniy Polyakov
2007-05-11  9:52                       ` Krishna Kumar2
2007-05-11  9:56                         ` Evgeniy Polyakov
2007-05-11 11:30                           ` jamal
2007-05-11 11:53                             ` Evgeniy Polyakov
2007-05-11 12:15                               ` jamal
2007-05-10 21:27       ` David Stevens
2007-05-10 21:40         ` David Miller
2007-05-10 21:50           ` Gagan Arneja
2007-05-10 22:06             ` David Miller
2007-05-11  9:46               ` Krishna Kumar2
2007-05-10 21:41         ` Eric Dumazet
2007-05-10 22:09           ` Rick Jones
2007-05-10 21:45         ` Rick Jones
2007-05-10 21:53           ` David Stevens
2007-05-10 20:21 ` Roland Dreier
2007-05-11  7:30   ` Krishna Kumar2
2007-05-11 11:21     ` Roland Dreier
2007-05-11  9:05 ` Andi Kleen
2007-05-11  8:32   ` Krishna Kumar2
2007-05-11  9:37     ` Andi Kleen
2007-05-11  8:50       ` Krishna Kumar2
2007-05-11 11:16       ` Roland Dreier
2007-05-13  6:00         ` Andi Kleen
2007-05-15 16:25           ` Roland Dreier
2007-05-15 20:18             ` David Miller
2007-05-15 20:52               ` Roland Dreier
2007-05-15 21:48                 ` Michael Chan
2007-05-15 21:08                   ` Roland Dreier
2007-05-15 22:05                     ` Michael Chan
2007-05-15 21:28                       ` David Miller
2007-05-18  7:04                         ` Michael Chan
  -- strict thread matches above, loose matches on Subject: below --
2007-05-11  7:14 Krishna Kumar2
     [not found] <OF0CAD6D87.DBE62968-ON872572DC.0073646A-882572DC.0073BEC2@us.ibm.com>
2007-05-15 21:17 ` Roland Dreier
     [not found]   ` <OFF5654BB8.74EC8DCB-ON872572DC.00752079-882572DC.00756B23@us.ibm.com>
2007-05-15 21:25     ` Roland Dreier
     [not found]       ` <OF21D475A2.5E5C88DE-ON872572DC.00763DE4-882572DC.00768A7E@us.ibm.com>
2007-05-15 21:38         ` David Miller
2007-05-15 21:32     ` David Miller
     [not found]       ` <OF6757F56D.EE5984FD-ON872572DC.0081026C-882572DC.00814B8F@us.ibm.com>
2007-05-15 23:36         ` David Miller
2007-05-21  7:56       ` Herbert Xu
     [not found]         ` <OF9ABCD08D.2CD1B193-ON872572E3.007A6FC1-882572E3.007ACE1A@us.ibm.com>
2007-05-22 22:36           ` David Miller
     [not found]             ` <OFCF3EB7F8.9740C0C7-ON872572E3.007DADF6-882572E3.007E0E7B@us.ibm.com>
2007-05-22 23:04               ` David Miller
2007-05-22 23:12             ` Herbert Xu

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=464375F9.1050600@gmail.com \
    --to=gaagaan@gmail.com \
    --cc=johnpol@2ka.mipt.ru \
    --cc=krkumar2@in.ibm.com \
    --cc=netdev@vger.kernel.org \
    --cc=rick.jones2@hp.com \
    --cc=sri@us.ibm.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).